Algorithms for Molecular Biology

Table 5 Estimates of AUC-ROC and AUC-PR of different methods for imbalanced dataset in predicting donor splice sites using human, bovine, fish and worm species

From: Identification of donor splice sites using support vector machine: a computational approach based on positional, compositional and dependency features

	Species	Approaches
	Species	MM1-SVM	LIK-SVM	WD-SVM	WDS-SVM	EFFECT	Proposed
AUC-ROC	Human	97.32	97.61	97.73	97.30	97.42	97.21
	Bovine	97.57	97.89	97.93	97.65	97.70	97.45
	Fish	97.71	97.85	97.92	97.77	97.57	97.41
	Worm	97.99	98.26	98.51	98.30	98.45	98.06
AUC-PR	Human	89.95	92.23	92.36	92.17	92.41	93.24
	Bovine	90.02	92.13	92.39	92.16	92.42	93.34
	Fish	90.10	92.18	92.43	92.26	92.47	93.38
	Worm	89.10	90.27	90.89	91.53	91.67	92.29

It can be seen that the values of AUC-ROC of proposed approach are at par with that of others, whereas the values of AUC-PR for the proposed approach are little higher than that of other approaches, in all the four species

Back to article page

ISSN: 1748-7188

Contact us

General enquiries: journalsubmissions@springernature.com